/VLM-ZSAD-Paper-Review

Reviews of papers on zero-shot anomaly detection using vision-Language models

VLM-ZSAD-Paper-Review

  • Vision-Language Models(VLMs)을 활용한 Zero-Shot Anomaly Detection(ZSAD)을 수행한 논문들을 리뷰합니다.
  • 리뷰 내용에 관해 수정해야하거나, 궁금한 부분 있으시다면 이메일(junyeong_son@korea.ac.kr)을 통해 연락 부탁드립니다.
  • [Youtube] 링크에는 서울대학교 산업공학과 DSBA 연구실 유튜브에서 직접 제작한 리뷰 영상을 포함시켰습니다.
  • [Github] 링크의 경우 official code가 아닐 수 있습니다.
Title Description Conference Year Review arXiv Github Youtube
WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentation WinCLIP CVPR 2023 [Review] [arXiv] [Github] --
AnoVL: Adapting Vision-Language Models for Unified Zero-shot Anomaly Localization AnoVL arXiv 2023 [Review] [arXiv] [Github] --
AnomalyCLIP: Object-agnostic Prompt Learning for Zero-shot Anomaly Detection AnomalyCLIP ICLR 2024 [Review] [arXiv] [Github] [Youtube]
PromptAD: Learning Prompts with only Normal Samples for Few-Shot Anomaly Detection PromptAD CVPR 2024 [Reivew] [arXiv] [Github] --
AdaCLIP: Adapting CLIP with Hybrid Learnable Prompts for Zero-Shot Anomaly Detection AdaCLIP ECCV 2024 [Review] [arXiv] [Github] [Youtube]
VCP-CLIP: A visual context prompting model for zero-shot anomaly segmentation VCP-CLIP ECCV 2024 [Review] [arXiv] [Github] --

To DO

  • AnomalyGPT: Detecting Industrial Anomalies using Large Vision-Language Models
  • CLIP-AD: A Language-Guided Staged Dual-Path Model for Zero-shot Anomaly Detection
  • FiLo: Zero-Shot Anomaly Detection by Fine-Grained Description and High-Quality Localization
  • Do LLMs Understand Visual Anomalies? Uncovering LLM Capabilities in Zero-shot Anomaly Detection
  • ClipSAM: CLIP and SAM Collaboration for Zero-Shot Anomaly Segmentation

Reference